INDEX

Explanations

news-related phrases or statements

oai_token-act-pair · gpt-3.5-turbo

instances of citations or references to sources

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

phony

-0.69

odox

-0.68

away

-0.65

 backward

-0.62

eez

-0.62

isSpecialOrderable

-0.60

athed

-0.59

eco

-0.59

ãĥ¥

-0.59

ific

-0.59

POSITIVE LOGITS

 "âĢ¦

0.79

 although

0.72

"[

0.71

 citing

0.68

 Kinnikuman

0.67

uthor

0.67

"(

0.66

Fib

0.66

 Mehran

0.65

 "...

0.64

Activations Density 0.156%

news-related phrases or statements

instances of citations or references to sources

No Comments

No Known Activations